CN103370910B - Utilize link aggregation to carry out method, system and the computer-readable medium of down hop convergent-divergent - Google Patents

Utilize link aggregation to carry out method, system and the computer-readable medium of down hop convergent-divergent Download PDF

Info

Publication number
CN103370910B
CN103370910B CN201180065318.2A CN201180065318A CN103370910B CN 103370910 B CN103370910 B CN 103370910B CN 201180065318 A CN201180065318 A CN 201180065318A CN 103370910 B CN103370910 B CN 103370910B
Authority
CN
China
Prior art keywords
grouping
described
down hop
packet
outlet
Prior art date
Application number
CN201180065318.2A
Other languages
Chinese (zh)
Other versions
CN103370910A (en
Inventor
D·B·格罗塞尔
E·J·瑞芙娜
Original Assignee
极进网络公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority to US41564310P priority Critical
Priority to US12/950,587 priority patent/US8660118B2/en
Priority to US61/415,643 priority
Priority to US12/950,587 priority
Priority to US12/972,391 priority patent/US8605726B2/en
Priority to US12/972,391 priority
Application filed by 极进网络公司 filed Critical 极进网络公司
Priority to PCT/US2011/061690 priority patent/WO2012068579A2/en
Publication of CN103370910A publication Critical patent/CN103370910A/en
Application granted granted Critical
Publication of CN103370910B publication Critical patent/CN103370910B/en

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00Routing or path finding of packets in data switching networks
    • H04L45/24Multipath
    • H04L45/245Link aggregation, e.g. trunking
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00Routing or path finding of packets in data switching networks
    • H04L45/60Router architecture
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00Routing or path finding of packets in data switching networks
    • H04L45/66Layer 2 routing, e.g. in Ethernet based MAN's
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00Routing or path finding of packets in data switching networks
    • H04L45/74Address processing for routing
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/25Routing or path finding through a switch fabric
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THIR OWN ENERGY USE
    • Y02D50/00Techniques for reducing energy consumption in wire-line communication networks
    • Y02D50/30Techniques for reducing energy consumption in wire-line communication networks by selective link activation in bundled links

Abstract

The theme of describing in literary composition comprises and utilizes link aggregation to carry out method, system and the computer-readable medium of down hop convergent-divergent. According to the theme of describing in literary composition scheme, provide a kind of system for down hop convergent-divergent. Described system comprises grouping converter, and it comprises multiple packet handlers of selecting operation for carrying out down hop and link aggregation group (LAG). In described multiple packet handlers, entrance packet handler is configured to: for the reception grouping with the down hop on different grouping processor, instruction need to export down hop and select operation. The outlet packet handler of described multiple packet handlers is configured to: for having indicated outlet down hop to select the grouping of operation, carry out outlet down hop and member and select operation, wherein, the reforwarding office of described grouping is limited to the movable LAG group membership of described outlet packet handler this locality.

Description

Utilize link aggregation to carry out method, system and the computer-readable medium of down hop convergent-divergent

Prioity claim

The application requires to enjoy the U.S. Provisional Patent Application sequence number of submitting on November 19th, 2010The U.S. Patent Application Serial Number 12/950,587 that on November 19th, 61/415,643,2010 submits to, withAnd the rights and interests of the U.S. Patent Application Serial Number 12/972,391 of submission on December 17th, 2010; At thisBy it, each one is incorporated to herein in full by reference.

Technical field

Theme described herein relates to grouping converter. More specifically, theme described herein relates toAnd utilize link aggregation to improve the grouping converter of down hop convergent-divergent.

Background technology

ARP(Address Resolution Protocol) is transmitting between equipment between IP address and MAC AddressThe agreement that binding is followed. In the time that equipment need to send IP grouping to another equipment, transmitting apparatus is necessaryFirst determine corresponding to down hop and (lead to next machine in the path of destination or at next machineFor being destination itself in the situation of destination) the MAC Address of IP address. Described transmitting apparatusCan determine next-hop mac address with ARP. Described distribution of machine will have next-hop IP groundThe ARP request of location is broadcast to all machines in special domain, and IP address packet is contained in this ARP requestMachine make and replying with the MAC Address of this machine. Once distribution of machine has obtained down hopMAC Address, described distribution of machine stores the binding between described IP address and MAC Address intoIn reservoir, thus can use described binding by forwarding of packets subsequently to uniform machinery.

In some forwarding of packets frameworks, separately by entrance packet handler, that is, and with reception groupingThe packet handler that is associated of the port of grouping converter, carry out and be used to received groupingDetermine searching of next hop information. Described entrance packet handler is carried out in the grouping based on receivedExecution level 3 route queryings of IP address, thereby be not only in described packet identification grouping converterDestination port is also described packet identification destination MAC Address. Utilize such flowing for the whole networkThe entrance of amount design is searched framework, because the grouping taking specific main frame as destination may be arrived at groupingThe arbitrary port of forwarding unit, thereby the memory being associated with each packet handler must be stored pinWhole ARP binding to the All hosts that is connected to described grouping converter is gathered.

Because the memory that can be accessed by each packet handler is normally limited, thereby can connectMay be subject to the limit of the memory that single packet handler can use to the quantity of the main frame of grouping converterSystem, this has greatly limited the scalability (scalability) of grouping converter. For example,, at each pointGroup processor all have fixed amount for storing the forwarding of packets framework of memory of ARP binding,The maximum quantity of arp table entry is subject to the storage quantitative limitation of single packet handler, and this is because everyOne packet handler must be stored identical whole ARP binding set. In other words, each packet transactionThe requirement that device is stored identical whole ARP binding set makes the scalability of system reduce factor1/ (quantity of packet handler). Once reach the ARP binding storage that can use by single packet handlerThe maximum quantity of the arp table entry of spatial limitation, increase so packet handler can not improve grouping turnThe quantity of the main frame that the equipment of sending out can be served. Such restriction is undesirable, especially divides at needsWhen group forwarding unit is served a large amount of connection main frames.

Utilizing when virtual, occur that a kind of hope makes grouping converter serve a large amount of connectionsThe specific network configurations of main frame. In virtualized environment, single physical main frame can dispose multiple voidIntend main frame. Each fictitious host computer has SOS, the network address and the MAC Address of himself.Each fictitious host computer is by the ARP binding needing in grouping converter. Thereby, by a lot of physical hostsBe connected to grouping converter, wherein each physical host comprises multiple fictitious host computers, will further carryThe high demand to scalable grouping converter framework.

Therefore, consider these difficulties, need to be used for method, system and the computer of down hop convergent-divergentComputer-readable recording medium. But, in addition, although improved down hop convergent-divergent can utilize server virtualization orThe client of large data center is helped have in the layer 2 forwarding territory that person is large, but these identical clients alsoOften seriously relying on link aggregation (being called again port load shares) restores and load balance.Link aggregation group (LAG) can be made up of the multiple physical ports that are equivalent to a group. Client is to LAGBehavior exist two kinds of expections, this be in any distributed IP down hop Zoom method that will develop allMust solve: (1) from multiple sources to particular mac address (and network address) (or from single sourceTo multiple addresses) grouping can utilize all active ports in LAG for load balance; WithAnd when (2) port downlink or uplink in LAG, described network adapts to realize expection (1) fast.

Therefore,, for developed any for for the method and system of down hop convergent-divergent, all mayWish that such exploitation can not hinder the ability of such method and system junction link polymerization.

Summary of the invention

The theme of describing in literary composition comprises and utilizes link aggregation to carry out method, system and the meter of down hop convergent-divergentCalculation machine computer-readable recording medium. According to the theme of describing in literary composition scheme, provide a kind of for down hopThe system of convergent-divergent. Described system comprises grouping converter, and it comprises for carrying out down hop and linkAggregation group (LAG) is selected multiple packet handlers of operation. In described multiple packet handlers,Entrance packet handler is configured to: for the grouping of arriving at from the outside port of grouping converter,Carry out down hop and LAG member and select operation. Also described entrance packet handler is configured to: forHave the grouping receiving of the down hop on different grouping processor, instruction need to export nextJump and select operation. The outlet packet handler of described multiple packet handlers is configured to: for instructionOutlet down hop select the grouping of operation, carry out outlet down hop and member and select operation, wherein,The reforwarding office of described grouping is limited to the movable LAG group membership of described outlet packet handler this locality.

Can utilize the non-instantaneous computer-readable medium with the executable instruction being stored thereon to realizeThe theme for down hop convergent-divergent that utilizes link aggregation of describing in literary composition, described instruction is by computerProcessor control described computer executed step while carrying out. For realizing showing of theme that literary composition describesPlasticity computer-readable medium can comprise chip memory devices, disk storage device, FPGAEquipment and special IC. In addition, realize the computer-readable medium of the theme of describing in literary composition passableBe positioned in individual equipment or computing platform, or can cross over multiple equipment or computing platform distribution.

Brief description of the drawings

The preferred embodiment of the theme of describing in literary composition is described referring now to accompanying drawing, wherein:

Fig. 1 is according to the block diagram of the grouping converter of the embodiment of the theme of describing in literary composition;

Fig. 2 shows the square frame of the table in the grouping converter that does not have modified link aggregationFigure;

Fig. 3 is the grouping with modified link aggregation according to the embodiment of the theme of describing in literary compositionThe block diagram of forwarding unit; And

Fig. 4 shows the down hop of utilizing link aggregation according to the embodiment of the theme of describing in literary compositionThe flow chart of the illustrative methods of convergent-divergent.

Detailed description of the invention

The theme of describing in literary composition comprise utilize link aggregation the method for down hop convergent-divergent, system andComputer-readable medium. Fig. 1 shows the link that utilizes according to the embodiment of the theme of describing in literary compositionThe block diagram of the exemplary grouping converter of down hop convergent-divergent is implemented in polymerization. With reference to figure 1, forwarding of packetsEquipment 100 can have layer 2 exchange capacity and layer 3 forwarding of packets ability, wherein, and layer 2 and layer 3Refer to the respective layer in osi protocol stack. Grouping converter 100 comprises multiple I/O (I/O)Module 102, each module has at least one port one 04, transmits grouping for the main frame to connectedOr receive grouping from it, described main frame is for example server, personal computer, router etc. OftenOne I/O module 102 comprises packet handler 106 and memory 108. Each packet handler 106 profitWith the forwarding of packets information and executing forwarding of packets operation being stored in its corresponding memory 108. InstituteIn the example illustrating, each memory 108 comprises: for the MAC of accumulation layer 2 forwarding of packets informationForwarding database 110, for floor 3 longest prefix match or the main frame road of accumulation layer 3 forwarding of packets informationBy table 112, for storing the next skip list 114 of outlet port, MAC Address and vlan information,And for storing trunking (trunk) table 120 of LAG information, described LAG information comprisesOutlet slot and the port information of LAG group.

Combination storage down hop binding (for example, the ARP of layer 3 routing table 112 and next skip list 114Binding). For example, in layer 3 routing table 112, carry out and search, to determine corresponding to the grouping receivingIn down hop identifier or the index of IP address, destination. Utilize described down hop identifier or indexIn next skip list 114, carry out and search, this next skip list 114 can be stored the ground corresponding to destination IPThe MAC Address of location and vlan information. Alternatively, if IP address, destination is connected to LAG,Searching in next skip list 114 can obtain opposite end mouth and converge the index of table 120 so. TrunkingTable 120 comprises the entry corresponding to LAG group, wherein, corresponding to the entry of LAG group comprise forLAG group membership's outlet slot and port information. Each packet handler 106 can be based on grouping ginsengNumerical value utilizes hashing algorithm to select in the middle of LAG group membership, and can be by described forwarding of packetsTo the packet handler being associated with described outlet slot and port. But, owing to using distributed ARPConvergent-divergent, also can be carried out forwarding of packets and searched by outlet packet handler, and in LAG groupThe hashing algorithm of the outlet packet transaction of selecting in member can not be equal to entrance packet handlerOn hashing algorithm, may appear at the convolution effect of multiple packet handler repeatings grouping. Wen ZhongThis convolution effect in the time selecting in LAG group membership avoided in the theme of describing, evenAlso like this in the time that execution is searched for the entrance and exit forwarding of packets of distributed ARP convergent-divergent, below willThis is given to more detailed description.

Except I/O module 102, grouping converter 100 comprises switching fabric 116, and it is responsible forBetween I/O module 102 and between I/O module 102 and switch management module 118, forwarding grouping.The overall operation of switch management module 118 control packet forwarding units 100. In addition exchange of management mould,Piece 118 can participate in IP routing protocol, with packed layer 3 routing tables 110, and can be at layer 3In routing table 112, packing object does not divide into groups to carry out the forwarding lookup of slow path.

Distributed ARP convergent-divergent

Utilize distributed ARP convergent-divergent, ARP binding can be crossed over packet handler and be distributed, thereby will leadThe ARP binding of machine is only stored in the next skip list of the packet handler that connects described main frame. Grouping placeIn the next skip list of reason device, can represent for the entry of other main frames that are not connected to this packet handlerFor the relaying entry in the next skip list of packet handler use. If packet handler (hereinafter referred to asEntrance packet handler) grouping that receives from outside port cause with next skip list relaying entryCoupling, is forwarded to without revising grouping place being associated with described outlet port by described grouping soReason device (hereinafter referred to outlet packet handler). Outlet packet handler will be in its L3 routing table andIn next skip list, carry out and search, determine the MAC Address of grouping, replace the MAC ground in described groupingLocation, and forward the grouping from grouping converter 100.

For example, the next skip list 114 in different I/O module 102 can be configured to comprise different underOne hop-information. For example, set next skip list 114 can be stored and be connected to and this next skip list 114The MAC Address of the main frame of the packet handler being associated, is connected to other packet handlers but do not compriseThe MAC Address of main frame. Selected next skip list 114 comprises and is connected to other packet transactionThe relaying entry of the main frame of device, wherein, in destination port identification outlet I/O module, conduct did not previously makeWith " mixing " port of port. Described Hybrid port can be any port for forwarding of packets not.

Receive while carrying out the grouping of relaying by entrance packet handler institute at outlet packet handlerState grouping and will comprise the fabric header (fabricheader) of instruction destination port. By entrance grouping placeIt is upper that fabric header is inserted into grouping by reason device, thereby switching fabric can be guided to described grouping justTrue outlet packet handler. If fabric header outlet port parameter instruction Hybrid port, so thisThe parameter of sample is carried out triggering outlet packet handler to search in its L3 routing table and next skip list, withDetermine the MAC Address of described grouping and the outlet port of described grouping. Afterwards, outlet packet transactionDevice is by the MAC Address of replacing in grouping, and forwarding is from the grouping of grouping converter 100.

It should be noted that and require the next skip list being associated with each packet handler only to store for systemIn the single relaying entry of each additional packet processor, and no matter be connected to other packet handlersThe quantity of main frame how, and this will greatly reduce each packet handler and be necessary for and be connected to other processorsThe data volume that must store of main frame. Shall also be noted that the L3 road that is associated with each packet handlerBe there are differences each other by table 112, to show the difference in corresponding next skip list. At 2010 11In the U.S. Patent Application Serial Number 12/950,587 that the moon 19 was submitted to, disclose and there is such ARPThe exemplary configuration of the grouping converter 100 of the distributed arrangement of binding, by reference by this applicationDisclosure be all incorporated to herein.

Owing to grouping converter 100 can being configured to comprise the distributed storage dress of ARP bindingPut, instead of be confined to the storage device that can be bound for storing ARP by single packet handler, because ofAnd the limited amount of the main frame that grouping converter 100 can connect is in crossing over all packet handlersCan be used for the storage total amount of ARP binding. Thereby, compared with forming with Typical Disposition, such configuration energyThe remarkable lifting of scalability is enough provided. In addition, because each packet handler is responsible for direct-connectedThe whole ARP binding of main frame, thus the quantity of the main frame that can serve of grouping converter 100 along withThe quantity of packet handler and convergent-divergent. Therefore, can improve by adding extra packet handlerThe connectedness of grouping converter 100.

Thereby, utilize with the relaying entry on outlet packet handler and search and combine with outlet down hopThe distributed storage of ARP binding, realized improved ARP convergent-divergent. Can utilize and to describe in literary compositionThe grouping converter with ARP convergent-divergent greatly improves the physics that this grouping converter can be servedAnd/or the quantity of fictitious host computer. For example, because packet handler distribution is crossed in the storage of down hop binding,Thereby the limited amount of the main frame that can serve is in total memory storage sky of crossing over all packet handlersBetween, and be not limited to the memory storage of single packet handler. For example,, if set forwarding of packetsEquipment has four packet handlers, and so described grouping converter can hold and has some entriesDistributed next skip list, described entry can be stored in the combination memory space of four packet handlersIn. The sum of the down hop entry that can store will can not equal a packet handler respective counts justAmount four times because relaying entry occupies some spaces on entrance packet handler. In addition main frame,Between packet handler, can not equally distributed. Thereby the theme of describing in literary composition makes groupingForwarding unit is more scalable, and this is to allow the more main frame of service because increase extra packet handlerReduce the possibility of the maximum quantity that reaches next skip list entry simultaneously.

Utilize the distributed ARP convergent-divergent of LAG

But, as mentioned above, also wish so improved ARP convergent-divergent to be attached to link is also providedPolymerization is for restoring in the middle of the system with load balance. Increasing link aggregation may mean, even pointGroup processor from outside port receive grouping cause with next skip list relaying entries match, stillIf described relaying entry is associated with LAG, so still must carries out LAG member and select operation.Fig. 2 is the more detailed diagram of grouping converter 100, and it shows in link aggregation configuration works asThe exemplary configuration of layer 3 routing table 112, next skip list 114 and the trunking table 120 of middle use.For the sake of simplicity, in Fig. 2, omitted the remaining part shown in Fig. 1. In Fig. 2, I/O module 102Be associated with the different cards slot in grouping converter 100. In the middle of shown configuration, singleI/O module is associated with slot 1, and single I/O module is associated with slot 8, and single I/O mouldPiece is associated with slot 9. For the object of this example, also suppose that each I/O module comprises single pointGroup processor. But each I/O module and/or slot use multiple packet handlers also should to be recognizedBe decided to be in the scope of the theme of describing in literary composition.

In an example shown, each port one 04 has the port numbers and the slot number that are associated. For example,1:48 represents slot 1, port 48, and 8:21 represents slot 8, port 21 etc. Slot 1: port 48Be connected to main frame 200, slot 8: port one is connected to main frame 202, and slot 9: port 48 connectsTo main frame 204. Similarly, all the other ports of grouping converter 100 connect together as LAG130Be connected to the main frame 206 of the IP address/MAC Address combination having shown in Fig. 2. Particularly, slot8: port 21, slot 8: port 22 and slot 9: each in port 23 connects via L2 switch 208Be connected to main frame 206. Main frame 200,202,204 and 206 can be physical equipment or physical equipmentCombination with virtual unit. For example, main frame 200,202,204 and 206 can be have or do not haveThe server of virtualization software, client computer, router etc.

As shown in Figure 2, LAG member selects by completing as follows: to each I/O module 102On trunking table 120 entry query intention for the main frame that is associated with LAG130 (for example,Main frame 206 in Fig. 2) all down hops. Particularly, for example, each trunking table 120Entry 0 comprise the list as each port one 04 of the member of LAG130. It should be noted thatIn this configuration, in trunking table 120, be associated with each I/O module 102 and packet handlerEntry is equal to, and wherein, each trunking table 120 comprises for as being connected to forwarding of packetsThe entry of the member's of the LAG130 of equipment 100 all of the port, and do not consider the physics that member connectsPort.

In the situation that there is no distributed IP ARP convergent-divergent, entrance packet handler utilizes relevant grouping ginsengIn the middle of the port of the hash of number from LAG130, select outlet port, outlet packet handler is not carried outAny ARP or LAG search. But, in the time implementing as described above distributed ARP convergent-divergent, enterThe two all plays a role mouth packet handler and outlet packet handler in forwarding of packets. Owing to relating toTwo packet handlers independently, thereby utilize conventional based on hash at two packet handler placesMay there is convolution problem mentioned above in LAG decision-making. As example, with reference to figure 2, can be byBe intended for and transfer to main frame 206 from the grouping of main frame 200. If the LAG behavior of the standard of employing,Entrance packet handler (packet handler on the I/O module 102a on slot 1) can be right soHash is carried out in grouping, and selects the outlet port 8:21 on slot 8. Exporting afterwards packet handler (insertsPacket handler on I/O module 102b on groove 8) carry out L3 and search, to forward described grouping,And again carry out hash. If hashing algorithm is in grouping place being associated with I/O module 102a and 102bBetween reason device, be not identical, this second hash can be selected the outlet on another different slot soPort, for example, the port 9:23 on slot 9, as long as new outlet port is also one of LAG130Part. For example, in (, the grouping on the I/O module 102c on slot 9 of the processor from finalProcessor) export and access intermediate packets processor (for example, the I/O module 102b on slot 8 beforeOn packet handler) this " convolution effect " of grouping (as being identified as path P 1 in Fig. 2Dotted line shown in) excessive use switching backplane bandwidth, be therefore do not wish occur.

For fear of such effect, a kind of selection can be by unique outlet port for each onlyIn the situation of one IP destination, trunking table is carried out to pre-programmed, to carry out a kind of accurate load balance.This configuration will be avoided " convolution effect ", but it is by the optimum load not allowing for a serverShare, because from all of the port on an entrance packet handler and taking a main frame as destinationGrouping at every turn all will be taking same port as outlet. Such scheme also requires on LAG, to keep with everyThe list of one main frame that is associated of outlet port, thereby no matter when in the time of LAG port upstream or downstreamAll need ten hundreds of trunking tables to upgrade, or no matter when when adding on LAG or deletingEach main frame of balance again when port. This layout will can not change fast reaction will be provided LAG.

Or, can be by system configuration for keep as shown in Figure 3 two LAG space in trunking table.For the front panel port arrangement of each LAG group utilize to(for) user, can automatically prepare and keep singleOnly special LAG group. This special LAG group can comprise front panel port one 04 or grouping placeReason device " Hybrid port " 104a. While no matter when front panel LAG group being made to renewal, can both be fastUpgrade described special LAG group fastly. If there is no movable LAG end on local packet handlerMouthful, described special LAG can be configured to so comprise the Hybrid port corresponding to each outlet portNumber. But, if a certain LAG port is movable on local packet handler, so can by described inSpecial LAG is configured to only comprise the front panel port numbers of local activity. This configuration makes LAG groupMaximum quantity reduce by half (for example, becoming 64 from 128), but it has met two kinds of expectations of customers: negativeCarry the grouping that shares to distributed IP arp entry, and to the fast reaction changing in LAG. ShouldWork as attention, only have distributed IP down hop host entry to utilize described special LAG group. Other forwardingsData base entries, for example layer 2MAC entry or multicast entry, can be used the LAG of " standard "Group, and be not subject to the impact of this method.

With regard to the specific implementation of special LAG group, Fig. 3 illustrates, each trunking table 120 exceptFor example have, outside the member's who comprises set LAG main entry (, entry 0), each port convergesPoly-table 120 also has the supplementary entry (for example, entry 64) being associated with corresponding special LAG.Just as different next skip list 114 can be different for each I/O module 102, each port convergesThe supplementary entry of poly-table 120 can be also different, to reflect for from set I/O module 102Which member of the set LAG of grouping forwarding is only the difference of preferred outlet port. Particularly,The supplementary entry of each trunking table 120 can only list as the member's of LAG port preferablySubset.

For example,, in the layout shown in Fig. 3, according to expection, from port one: the main frame 200 48 toThe grouping of main frame 206 using cross over these ports as the member of LAG (for example, 8:21,8:22 withAnd 9:23) distribute. If any port in these ports is descending, so described grouping will be crossed overRemaining active port distributes. In addition, the supplementary entry of trunking table 120 can list corresponding toEach LAG member's Hybrid port number, instead of accurately list each port. For example, due toIn LAG, there is no any active port of main frame 200 this locality, thereby the supplementary bar of trunking table 120aOrder only comprises the Hybrid port number (for example, 8:49 and 9:49) corresponding to outlet port. By this sideFormula, entrance packet handler only need identify suitable outlet packet handler, instead of selects accuratelyOutlet port.

In addition, the in the situation that of comprising multiple LAG member in single I/O module, can be by correspondenceThe Multi-instance of Hybrid port number is listed as in the supplementary entry of trunking table 120a, is transmitted to keepingThe equal distribution of each member's of LAG grouping. For example, the entry 64 of trunking table 120a canWith two examples that comprise Hybrid port 8:49, this is because described Hybrid port number and outlet port8:21 and 8:22 are associated. As a result, may export port according to what be associated with each corresponding module2 to 1 ratios, it is to be forwarded to I/O module 102c that the grouping that is forwarded to I/O module 102b will be tended toTwo times of grouping.

On the contrary, for from the main frame 202 port 8:1 to the grouping of main frame 206, preferablyOnly select the local port (, 8:21 and 8:22) on same packet handler. Correspondingly, from portConverge in the supplementary entry of showing 120b and omit non-local port (, 9:23). Similarly, from portMain frame 204 on 9:48 guides to the grouping of main frame 206 can be preferably always with local port (being 9:23)For outlet. Only at port 9:23 when descending, the grouping from main frame 204 to main frame 206 just can be acrossMore port 8:21 and 8:22 distribute, and this is because protect without any local port on this packet handlerHold activity.

Fig. 4 show according to the embodiment of the theme of describing in literary composition utilize link aggregation underOne jumps the flow chart of the example process of convergent-divergent. With reference to figure 4, in step 400, next hop information existsAnd between the memory that the different grouping processor in grouping converter is associated, distribute, thus make withThe different down hop binding of memory stores that different grouping processor is associated. For example,, as Fig. 3 instituteShow, (or the port of packet handler reference converges the next skip list being associated with each packet handlerPoly-table) can store the complete MAC Address of direct-connected main frame, and can store and be connected to itThe relaying entry of the main frame of his packet handler.

In step 402, receive grouping at entrance packet handler place. Described grouping has and LAGThe down hop being associated. For example, refer again to Fig. 3, at slot 1: the grouping that port 48 places receiveCan there is IP address, destination 23.1.1.89, its have corresponding to be connected to I/O module 102b andThe down hop of both LAG of 102c.

Return to Fig. 4, in step 404, for the entrance being associated with the active port member of LAGPacket handler, carries out ARP and search to determine destination MAC Address, and forwarding turns from groupingThe grouping of the equipment of sending out. Particularly, in the middle of the member of the LAG that is associated with entrance packet handlerMAC Address selects your destination.

Return to Fig. 4, in step 406, for unconnected the entering of active port member of LAGMouthful packet handler, first by described packet relay to exporting packet handler. Continue current example,For arriving at slot 1: the grouping that is addressed to 23.1.1.89 of port 48, LAG member selects to operate willBe chosen in one of Hybrid port to listing in the supplementary entry of trunking table 120a number, and based on thisOne selects to forward the packet to slot 8 packet handlers or slot 9 packet handlers. DescribedRelay for packets will have previous untapped port numbers (, Hybrid port) in its fabric header.

Return to Fig. 4, in step 408, in outlet, packet handler place receives grouping, described structure reportHybrid port address in head is triggered ARP and is searched, and to determine destination MAC Address, and forwardsFrom the grouping of described grouping converter. Particularly, the LAG from being associated with outlet packet handlerMember in the middle of the MAC Address that selects your destination. Continue present example, if will be addressed to23.1.1.89 forwarding of packets is to slot 8, and the 2nd LAG member selects operation converging to port soThe slot 8 of listing in the supplementary entry of poly-table 120b: port 21 or slot 8: select in the middle of port 22Select. To use MAC Address and VLAN mark corresponding to selected outlet port to replace described pointMAC Address in group and VLAN mark. To utilize selected port from grouping converter 100Forward grouping.

Although above-described theme relates to distributed ARP convergent-divergent, the theme of describing in literary composition is not limitIn ARP convergent-divergent, in the case of the scope of the theme described in not deviating from literary composition, can be applied toThe down hop convergent-divergent of any kind. For example, in the case of the scope of the theme described in not deviating from literary composition,The theme of describing in literary composition can be found to the distributed of binding (being equivalent to the ARP of IPv6) for neighboursStorage.

Should be appreciated that the scope in the case of not deviating from current disclosed theme and can change current public affairsThe various details of the theme of opening. In addition, above stated specification is for illustrational object, instead ofFor the object of restriction.

Claims (25)

1. utilize link aggregation to carry out a system for down hop convergent-divergent, described system comprises:
Grouping converter, comprising:
Multiple packet handlers, select operation for carrying out down hop and link aggregation group LAG;
Wherein, the entrance packet handler in described multiple packet handlers is configured to: forThe grouping of arriving at from the outside port of described grouping converter, carries out link aggregation group LAG member choosingSelect operation, and described entrance packet handler is configured to: for having in different grouping processingThe grouping receiving of the down hop on device, instruction need to export down hop and select operation; And
Wherein, the outlet packet handler in described multiple packet handlers is configured to: forIndicate outlet down hop to select the grouping of operation, carried out outlet down hop and member and select operation, andThe forwarding of described grouping is limited to the movable LAG group membership of described outlet packet handler this locality.
2. system according to claim 1, wherein, with depositing that each packet handler is associatedReservoir storage is used for the relaying entry of the down hop that is connected to all the other packet handlers.
3. system according to claim 2, wherein, with depositing that each packet handler is associatedReservoir storage is for each single relaying entry of all the other packet handlers.
4. system according to claim 2, wherein, with the relaying bar on entrance packet handlerThe reception of the grouping of order coupling triggers outlet packet handler execution outlet down hop and searches.
5. system according to claim 3, wherein, described relaying entry is not stored and is connected to itThe MAC Address of the main frame of remaining packet handler.
6. system according to claim 1, wherein, with depositing that each packet handler is associatedReservoir is only for to be connected directly to the down hop store M AC address of each packet handler, thereby will be used forBetween the memory that down hop binding is distributed in and all packet handlers are associated of All hosts.
7. system according to claim 1, wherein, each memory stores: layer 3 main frame orPerson's layer 3 longest prefix match routing table, IP address information is mapped to next hop information by it; And underOne jumps table, and described next hop information is mapped to layer 2 forwarding of packets information by it.
8. system according to claim 7, wherein, for layer 3 main frame or longest prefix matchRouting table and described next skip list can with memory share.
9. system according to claim 8, wherein, each memory stores MAC forwards numberAccording to storehouse, described MAC forwarding data library storage is connected to the MAC of the main frame of described grouping converterAnd vlan information.
10. system according to claim 1, wherein, is associated with described packet handlerMemory stores is for the down hop binding of actual host and fictitious host computer.
11. systems according to claim 1, wherein, down hop binding comprises address resolution associationView ARP binding.
12. systems according to claim 1, wherein, down hop binding comprises that IPv6 neighbours send outNow binding.
13. 1 kinds of methods for down hop convergent-divergent, described method comprises:
Next hop information is distributed in and comprises different in the grouping converter of multiple packet handlersIn the memory that packet handler is associated, thereby make the memory being associated with different grouping processorStore different down hop bindings;
Receive grouping at entrance packet handler place, described grouping has relevant to link aggregation group LAGThe down hop of connection;
For the grouping of arriving at from the outside port of described grouping converter, carry out link aggregation groupLAG member selects operation;
For the grouping receiving with the down hop on different grouping processor, instruction need to go outMouthful down hop is selected operation, and by described packet relay to exporting packet handler; And
For having indicated outlet down hop to select the grouping of operation, carry out outlet down hop and LAG memberSelect operation, wherein, described LAG member selects operation to go out described in the forwarding of described grouping is limited toThe movable LAG group membership of mouth packet handler this locality.
14. methods according to claim 13, wherein, are associated with each packet handlerMemory stores is used for the relaying entry of the down hop that is connected to all the other packet handlers.
15. methods according to claim 14, wherein, are associated with each packet handlerMemory stores is for each single relaying entry of all the other packet handlers.
16. methods according to claim 14, wherein, with the relaying on entrance packet handlerThe reception of the grouping of entries match triggers outlet packet handler execution outlet down hop and searches.
17. methods according to claim 15, wherein, described relaying entry is not stored and is connected toThe MAC Address of the main frame of all the other packet handlers.
18. methods according to claim 13, wherein, are associated with each packet handlerMemory is only for to be connected directly to the down hop store M AC address of each packet handler, thereby will useBetween the memory that down hop binding is distributed in and all packet handlers are associated of All hosts.
19. methods according to claim 13, wherein, each memory stores: layer 3 main frameOr longest prefix match routing table, IP address information is mapped to next hop information by it; And down hopTable, described next hop information is mapped to layer 2 forwarding of packets information by it.
20. methods according to claim 19, wherein, for described layer 3 main frame or the longest beforeSew coupling routing table and described next skip list can with memory share.
21. methods according to claim 20, wherein, each memory stores MAC forwardsDatabase, described MAC forwarding data library storage is connected to the main frame of described grouping converterMAC and vlan information.
22. methods according to claim 13, wherein, described down hop binding comprises address solutionAnalyse agreement ARP binding.
23. methods according to claim 13, wherein, described down hop binding comprises IPv6Neighbours find binding.
24. 1 kinds of devices for down hop convergent-divergent, described device comprises:
For next hop information being distributed in and the grouping converter that comprises multiple packet handlersIn the memory that different grouping processor is associated, thereby make and depositing that different grouping processor is associatedReservoir is stored the unit of different down hop bindings;
For receive the unit of grouping at entrance packet handler place, described grouping has and link aggregationThe down hop that group LAG is associated;
For for the grouping of arriving at from the outside port of described grouping converter, carry out link aggregationGroup LAG member selects the unit of operation;
For for the grouping receiving with the down hop on different grouping processor, indicate and needExport down hop select operation, and by described packet relay to the unit that exports packet handler; WithAnd
For for having indicated outlet down hop to select the grouping of operation, carry out outlet down hop and LAGMember selects the unit of operation, and wherein, described LAG member selects operation by the forwarding limit of described groupingBe formed on the movable LAG group membership of described outlet packet handler this locality.
25. 1 kinds of systems for address down hop convergent-divergent, described system comprises:
Grouping converter, comprising:
For carrying out multiple packet handlers of forwarding of packets operation; And
With the memory that each packet handler is associated, described memory is used for main for storageThe down hop binding of machine, wherein, by entrance packet handler with relevant to described entrance packet handlerThe memory of connection is configured to down hop to be connected to the outlet grouping separating with described entrance packet handlerThe packet relay of processor is described outlet packet handler extremely, and wherein, will receive relay for packetsOutlet packet handler and memory thereof be configured to carry out down hop and search, thereby carry for described groupingFor destination MAC Address, and forward the packet to described down hop.
CN201180065318.2A 2010-11-19 2011-11-21 Utilize link aggregation to carry out method, system and the computer-readable medium of down hop convergent-divergent CN103370910B (en)

Priority Applications (7)

Application Number Priority Date Filing Date Title
US41564310P true 2010-11-19 2010-11-19
US12/950,587 US8660118B2 (en) 2010-11-19 2010-11-19 Methods, systems, and computer readable media for next hop scaling
US61/415,643 2010-11-19
US12/950,587 2010-11-19
US12/972,391 US8605726B2 (en) 2010-11-19 2010-12-17 Methods, systems, and computer readable media for next hop scaling with link aggregation
US12/972,391 2010-12-17
PCT/US2011/061690 WO2012068579A2 (en) 2010-11-19 2011-11-21 Methods, systems, and computer readable media for next hop scaling with link aggregation

Publications (2)

Publication Number Publication Date
CN103370910A CN103370910A (en) 2013-10-23
CN103370910B true CN103370910B (en) 2016-05-18

Family

ID=46084698

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201180065318.2A CN103370910B (en) 2010-11-19 2011-11-21 Utilize link aggregation to carry out method, system and the computer-readable medium of down hop convergent-divergent

Country Status (3)

Country Link
EP (1) EP2641360B1 (en)
CN (1) CN103370910B (en)
WO (1) WO2012068579A2 (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8605726B2 (en) 2010-11-19 2013-12-10 Extreme Networks, Inc. Methods, systems, and computer readable media for next hop scaling with link aggregation
US8660118B2 (en) 2010-11-19 2014-02-25 Extreme Networks, Inc. Methods, systems, and computer readable media for next hop scaling

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6192051B1 (en) * 1999-02-26 2001-02-20 Redstone Communications, Inc. Network router search engine using compressed tree forwarding table
CN1369991A (en) * 2002-04-01 2002-09-18 北京港湾网络有限公司 Dividing method for bond ports of switch and switch chip
CN1725761A (en) * 2005-06-10 2006-01-25 杭州华为三康技术有限公司 Method of implementing link aggregation of network equipment

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7190696B1 (en) 2002-03-05 2007-03-13 Force10 Networks, Inc. System and method for distributing packets among a plurality of paths to a destination
US8009668B2 (en) 2004-08-17 2011-08-30 Hewlett-Packard Development Company, L.P. Method and apparatus for router aggregation
JP4457058B2 (en) * 2005-08-26 2010-04-28 アラクサラネットワークス株式会社 Packet transfer device with filtering
US7593400B2 (en) * 2006-05-19 2009-09-22 Corrigent Systems Ltd. MAC address learning in a distributed bridge
JP4830942B2 (en) * 2007-03-28 2011-12-07 日本電気株式会社 Communication apparatus, communication system, and LAG management table management method used therefor

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6192051B1 (en) * 1999-02-26 2001-02-20 Redstone Communications, Inc. Network router search engine using compressed tree forwarding table
CN1369991A (en) * 2002-04-01 2002-09-18 北京港湾网络有限公司 Dividing method for bond ports of switch and switch chip
CN1725761A (en) * 2005-06-10 2006-01-25 杭州华为三康技术有限公司 Method of implementing link aggregation of network equipment

Also Published As

Publication number Publication date
EP2641360B1 (en) 2018-03-21
WO2012068579A3 (en) 2012-10-04
WO2012068579A2 (en) 2012-05-24
EP2641360A4 (en) 2016-11-02
CN103370910A (en) 2013-10-23
EP2641360A2 (en) 2013-09-25

Similar Documents

Publication Publication Date Title
US10547544B2 (en) Network fabric overlay
CN104426681B (en) The distributed multicast of endpoint
US10250529B2 (en) Systems and methods for performing logical network forwarding using a controller
CN105706398B (en) The method and system that virtual port channel in overlapping network rebounds
AU2012312587B2 (en) System and methods for controlling network traffic through virtual switches
KR101623416B1 (en) Communication method and system
US9270566B2 (en) Monitoring server load balancing
CN104584491B (en) Distributed virtual route and the system and method for exchanging (DVRS) are provided
US9325619B2 (en) System and method for using virtual lanes to alleviate congestion in a fat-tree topology
US8576853B2 (en) Two-layer switch apparatus avoiding first layer inter-switch traffic in steering packets through the apparatus
US8923155B2 (en) L3 gateway for VXLAN
CN103259727B (en) A kind of message forwarding method and equipment
US9264382B2 (en) System and method for routing traffic between distinct infiniband subnets based on fat-tree routing
EP2670090B1 (en) Multi-homing in an extended bridge
TWI434551B (en) Driven multicast traffic distribution on link-aggregate-group
Korilis et al. Capacity allocation under noncooperative routing
EP2859695B1 (en) System and method for layer-2 multicast multipathing
JP6129928B2 (en) Agile Data Center Network Architecture
EP2804350B1 (en) Method and apparatus for implementing and managing virtual switches
US8676980B2 (en) Distributed load balancer in a virtual machine environment
EP3210345B1 (en) Transparent network service header path proxies
CN102158386B (en) Distributed load balance for system management program
US8855116B2 (en) Virtual local area network state processing in a layer 2 ethernet switch
CN100421380C (en) Data mirroring in a service
US9614759B2 (en) Systems and methods for providing anycast MAC addressing in an information handling system

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant